Definition
Chain-of-Thought is a prompting technique that encourages the AI model to break down complex problems into intermediate reasoning steps before arriving at a final answer. By "thinking step by step," the model produces more accurate results on tasks requiring logic, math, or multi-step reasoning.
Example
Instead of directly answering "How many support tickets did we resolve last month?", the agent reasons: "First, I need to count resolved tickets. Then I need to filter by last month. Then I need to verify the count."
Why it matters
Chain-of-Thought reasoning reduces errors by 15-30% on complex tasks according to research from Google DeepMind. It's the difference between a model that guesses and a model that thinks.